Multi-pronged neuromodulation intervention engages the residual motor circuitry to facilitate walking in a rat model of spinal cord injury

Abstract

A spinal cord injury usually spares some components of the locomotor circuitry. Deep brain stimulation (DBS) of the midbrain locomotor region and epidural electrical stimulation of the lumbar spinal cord (EES) are being used to tap into this spared circuitry to enable locomotion in humans with spinal cord injury. While appealing, the potential synergy between DBS and EES remains unknown. Here, we report the synergistic facilitation of locomotion when DBS is combined with EES in a rat model of severe contusion spinal cord injury leading to leg paralysis. However, this synergy requires high amplitudes of DBS, which triggers forced locomotion associated with stress responses. To suppress these undesired responses, we link DBS to the intention to walk, decoded from cortical activity using a robust, rapidly calibrated unsupervised learning algorithm. This contingency amplifies the supraspinal descending command while empowering the rats into volitional walking. However, the resulting improvements may not outweigh the complex technological framework necessary to establish viable therapeutic conditions.

Fig. 1. Targeted deep brain stimulation of the midbrain locomotor region (MLR).

a Custom-made and commercial electrodes. Post-hoc anatomical evaluation of the electrode placement and localization of cholinergic (ChAT) neurons in the pedunculopontine nucleus (PPN), in the vicinity of the DBS electrode implantation site. PrCnF = Pre-Cuneiform Nuclei. Scales bar:, 100 μm (50 μm for inset). Repeated in n = 4 rats with similar results. b Locomotor speed modulation by MLR DBS as a response to different stimulation amplitude, frequency, and pulse width. Latency of locomotion initiation onset as a response to changing MLR DBS parameters (representative rat, n = 3 independent runs. Bars, data mean.). c Post-mortem reconstruction of midbrain tissues revealed that DBS induced a pronounced c-Fos expression in the vicinity of the electrode tips. Repeated in n = 4 rats with similar results in each (all shown). Scales bars: Left panel: 1mm, c-Fos 50 μm (inset 10 μm); Right panel 200 μm.

Fig. 2. Impact of serotonin agonists, EES and DBS to enable stepping.

a Representative photograph of the lesion epicenter, and graphical representation of sections with maximal damage for each rat from this group (n = 6 rats, all shown). Scale bar, 250 μm. The percentage of spared tissue is reported in each section. b Rats are tested at two weeks post-injury during stepping on a treadmill while supported in a robotic body weight support system. c Representative stick diagram decomposition of leg movements, oscillation of the whole limb (virtual limb linking the hip to the toe) and EMG activity of ankle muscles recorded at 2 weeks post-SCI with and without DBS. d Bar plots reporting mean values (n = 6 rats) of step height for each condition of stimulation. *, P < 0.05 using paired, one-tailed t-test (no 5HT, no stim vs 5HT, no stim, p = 0.043; 5HT, no stim vs 5HT + EES, p = 0.023; 5HT + EES vs 5HT, EES + MLR, p = 0.025); ns, not statistically significant. Bar plots, mean ± s.d. e Relationship between lesion size and increase in step height in the presence of both continuous EES + MLR and serotonergic agonists.

Fig. 3. Rehabilitative training improves the impact of DBS on locomotor performance.

a Timeline summarizing the experiments. b Rats are tested at 8 weeks post-injury during stepping on a treadmill while supported in a robotic body weight support system. c Representative stick diagram decomposition of leg movements, oscillation of the whole limb (virtual limb linking the hip to the toe) and EMG activity of ankle muscles recorded at 5 weeks post-SCI with 5HT agonists + EES versus 5HT agonists + EES + DBS. d Bar plots reporting mean values (n = 10 rats over 2 independent experiments) of step height for each condition of stimulation. *, P < 0.05, ** P < 0.01, ***, P < 0.001 using paired, one-tailed t-test (no 5HT, no stim vs 5HT, no stim, p = 0.009; no 5HT, no stim vs no 5HT, EES, p = 0.0004; 5HT, no stim vs 5HT + EES, p = 0.0046; 5HT + EES vs 5HT, EES + MLR, p = 0.0004). Error bars, s.d. e Principal components (PC) analysis (n = 5 rats). Gait patterns are displayed in the new reference frame created by PC1-2. The lines interpolate dots quantifying locomotor performance for the same rat in each of the four experimental conditions: EES and different levels of DBS intensities. The bar graphs report the scores on PC1 across conditions, which captured the enhancement of locomotor performance with DBS (gait patterns move in the direction of steps without injury). *, P < 0.05, ** P < 0.01, ***, P < 0.001 using paired, one-tailed t-test (0–100%, p = 0.0004; 33–100%, p = 0.004; 66–100%, p = 0.011). f Bar plots reporting mean values (n = 5 rats) of step height (0–100%, p = 0.0004; 33–100%, p = 0.004; 66–100%, p = 0.011) and hip excursion (0–100%, p = 0.0016; 33–100%, p = 0.0087; 66–100%, p = 0.039). The bar plot on the right reports the relative increase in the amplitude of the oscillations of each lower limb segments compared to EES only (Thigh-leg, p = 0.011; Thigh-foot, p = 0.0017). Paired, one-tailed t-tests. *, P < 0.05; **, P < 0.01; ***, P < 0.001. Bar diagrams, mean ± s.d.

Fig. 4. DBS improves locomotor performance during overground locomotion, despite stress responses.

a Schematic representation of the testing conditions: bipedal locomotion along a runway with tailored robotic assistance. All the evaluations are performed with EES and 5HT agonists. b Same representations are in Fig. 2 during overground locomotion. c Bar plots reporting the mean values (±s.d) of step height under the different conditions (No stim vs EES, n = 9 rats, p = 0.025; No stim vs MLR, n = 6 rats; EES vs EES + MLR, n = 11 rats, p = 0.0001 using paired, one-tailed t-test). d Quantification of stress responses using the Rat Grimace Scale, wherein grimaces are scored from 0 to 2, as exemplified on three photos of the same rat, from Sotocinal S et al. Mol Pain 2011 (CC-BY license). e Bar plots reporting the mean values of stress responses with and without DBS (p = 0.00086 paired, one-tailed t-test). f Bar plots reporting the mean values of latency and variability (p = 0.0156 and p = 0.0156 paired, one-tailed Wilcoxon signed-rank test) in the latency between the beginning of the trial and the initiation of locomotion with and without DBS (n = 6 rats). *, P < 0.05; **, P < 0.01; ***, P < 0.001. ns, not statistically significant. Bar diagrams, mean ± s.d. g Top: Distribution of c-Fos+ neurons in the midbrain following DBS stimulation. Each color refers to a different rat (n = 4 rats). The plot shows the relationships between the number of c-Fos+ neurons and the distance from the tip of the electrode for each rat.

Fig. 5. Encoding of locomotion in motor cortex and midbrain neurons in healthy rats.

a Experimental setup depicting neuronal recordings from right MI (32 channels) and left MLR (16 channels) regions, together with whole body kinematics and muscle activity. b 3D visualization of MI and MLR electrode locations, combining in all experimental rats (n = 6 rats). A-P, anteroposterior; D-V, dorso-ventral. c Stick diagram decomposition of left hindlimb movement together with EMG activity of ankle muscles and multiunit activity recorded from MI and MLR regions. d Representative channel recorded from the electrodes located in MI and MLR with locomotor-related modulation of activity. The raster displays 10 successive trials. Shaded area, s.e.m. e Normalized averaged activity from all locomotor-related MI and MLR multi-units during gait initiation. Shaded area, s.e.m. f Distribution of MI and MLR channels encoding the speed of locomotion based on their relative correlation (R²) with average speed (n = 7 rats).

Fig. 6. Encoding of locomotion in motor cortex and midbrain neurons in rats with SCI.

a Sequence of photo showing quadrupedal locomotion after SCI, together with a stick diagram decomposition of left hindlimb movement, EMG activity of leg muscles and multiunit activity recorded from MI and MLR regions. (b) and (c) show the same representations as in Fig. 4 for recorded performed 2–3 weeks after the contusion SCI. Shaded areas, s.e.m. d Distribution of MI and MLR channels encoding the speed of locomotion based on their relative correlation (R²) with average speed (n = 6 rats). e Bar plot reporting the relative number of multi-units encoding the speed of locomotion in intact rats and after SCI (n = 6 rats, p = 0.031 paired, one-tailed Wilcoxon signed-rank test). f Bar plots reporting the mean values of the time at which the firing rates of all locomotor-related multi-units from MI and MLR crossed a threshold of 2 standard deviations above the baseline. The time is shown for all rats, before (n = 7 rats, p = 0.0001) and after SCI (n = 6, p = 0.00066). One-tailed Wilcoxon signed-rank test. *, P < 0.05; **, P < 0.01; ***, P < 0.001. Bar diagrams, mean ± s.d.

Fig. 7. Neuroprosthetic system enabling volitional control of neuromodulation procedures.

Recording of multiunit activity from MI is transformed into a synthetic variable that combines the firing of all locomotor-related multiunit, named cumulative firing. The cumulative firing is fed to an unsupervised algorithm that decodes walk and idle stages to turn MLR stimulation on and off. EES is continuously applied to the lumbosacral spinal cord in the presence of 5-HT agonists. Rats can initiate and sustain bipedal locomotion with a tailored gravity-assist, supported in a robotic body weight support.

Fig. 8. Brain-controlled MLR DBS improves locomotor performance while alleviating stress responses.

a Same representation as in Fig. 5 during locomotion without DBS and with brain-controlled MLR DBS (M1-MLR). b Relationship between lesion size and increase in step height during brain-controlled MLR DBS. c Bar plots reporting the mean values of the Euclidian distance in the PC space between the different experimental groups (n = 4 rats, with additional n = 2 rats which did not recover voluntarily locomotion separately represented; p = 0.027, paired, one-tailed t-test). d Bar plots reporting mean values of step height and hip excursion (p = 0.0042 and p = 0.0088, paired, one-tailed t-test) without and with brain-controlled MLR stimulation (n = 4 rats, with additional n = 2 rats separately superimposed). e Bar plots reporting the mean values (n = 6 rats) of latency (MLR OFF vs ON, p = 0.016; M1-MLR vs MLR ON, p = 0.016; paired, one-tailed Wilcoxon signed-rank test) and variability (MLR OFF vs ON, p = 0.016; M1-MLR vs MLR ON, p = 0.016; paired, one-tailed Wilcoxon signed-rank test) in the latency between the beginning of the trial and the initiation of locomotion without MLR stimulation, with MLR DBS and with brain-controlled MLR DBS, and (f) associated mean values of stress responses under the same conditions (M1-MLR vs MLR ON, p = 0.043; paired, one-tailed t-test). *, P < 0.05; **, P < 0.01; ***, P < 0.001. Bar diagrams, mean ± s.d.